* [PATCH bpf-next v2 0/3] net: skbuff: skb bitfield compaction - bpf
@ 2023-03-21 1:41 Jakub Kicinski
2023-03-21 1:41 ` [PATCH bpf-next v2 1/3] net: skbuff: rename __pkt_vlan_present_offset to __mono_tc_offset Jakub Kicinski
` (3 more replies)
0 siblings, 4 replies; 5+ messages in thread
From: Jakub Kicinski @ 2023-03-21 1:41 UTC (permalink / raw)
To: martin.lau
Cc: ast, daniel, bpf, netdev, davem, edumazet, pabeni, Jakub Kicinski
I'm trying to make more of the sk_buff bits optional.
Move the BPF-accessed bits a little - because they must
be at coding-time-constant offsets they must precede any
optional bit. While at it clean up the naming a bit.
v1: https://lore.kernel.org/all/20230308003159.441580-1-kuba@kernel.org/
Jakub Kicinski (3):
net: skbuff: rename __pkt_vlan_present_offset to __mono_tc_offset
net: skbuff: reorder bytes 2 and 3 of the bitfield
net: skbuff: move the fields BPF cares about directly next to the
offset marker
include/linux/skbuff.h | 36 +++++++++----------
net/core/filter.c | 8 ++---
.../selftests/bpf/prog_tests/ctx_rewrite.c | 14 ++++----
3 files changed, 29 insertions(+), 29 deletions(-)
--
2.39.2
^ permalink raw reply [flat|nested] 5+ messages in thread
* [PATCH bpf-next v2 1/3] net: skbuff: rename __pkt_vlan_present_offset to __mono_tc_offset
2023-03-21 1:41 [PATCH bpf-next v2 0/3] net: skbuff: skb bitfield compaction - bpf Jakub Kicinski
@ 2023-03-21 1:41 ` Jakub Kicinski
2023-03-21 1:41 ` [PATCH bpf-next v2 2/3] net: skbuff: reorder bytes 2 and 3 of the bitfield Jakub Kicinski
` (2 subsequent siblings)
3 siblings, 0 replies; 5+ messages in thread
From: Jakub Kicinski @ 2023-03-21 1:41 UTC (permalink / raw)
To: martin.lau
Cc: ast, daniel, bpf, netdev, davem, edumazet, pabeni, Jakub Kicinski
vlan_present is gone since
commit 354259fa73e2 ("net: remove skb->vlan_present")
rename the offset field to what BPF is currently looking
for in this byte - mono_delivery_time and tc_at_ingress.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
include/linux/skbuff.h | 4 ++--
net/core/filter.c | 8 ++++----
tools/testing/selftests/bpf/prog_tests/ctx_rewrite.c | 6 +++---
3 files changed, 9 insertions(+), 9 deletions(-)
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index 3f3a2a82a86b..5a63878a4550 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -955,7 +955,7 @@ struct sk_buff {
__u8 csum_valid:1;
/* private: */
- __u8 __pkt_vlan_present_offset[0];
+ __u8 __mono_tc_offset[0];
/* public: */
__u8 remcsum_offload:1;
__u8 csum_complete_sw:1;
@@ -1078,7 +1078,7 @@ struct sk_buff {
#define TC_AT_INGRESS_MASK (1 << 7)
#define SKB_MONO_DELIVERY_TIME_MASK (1 << 5)
#endif
-#define PKT_VLAN_PRESENT_OFFSET offsetof(struct sk_buff, __pkt_vlan_present_offset)
+#define SKB_BF_MONO_TC_OFFSET offsetof(struct sk_buff, __mono_tc_offset)
#ifdef __KERNEL__
/*
diff --git a/net/core/filter.c b/net/core/filter.c
index 50f649f1b4a9..3370efad1dda 100644
--- a/net/core/filter.c
+++ b/net/core/filter.c
@@ -9185,7 +9185,7 @@ static struct bpf_insn *bpf_convert_tstamp_type_read(const struct bpf_insn *si,
__u8 tmp_reg = BPF_REG_AX;
*insn++ = BPF_LDX_MEM(BPF_B, tmp_reg, skb_reg,
- PKT_VLAN_PRESENT_OFFSET);
+ SKB_BF_MONO_TC_OFFSET);
*insn++ = BPF_JMP32_IMM(BPF_JSET, tmp_reg,
SKB_MONO_DELIVERY_TIME_MASK, 2);
*insn++ = BPF_MOV32_IMM(value_reg, BPF_SKB_TSTAMP_UNSPEC);
@@ -9232,7 +9232,7 @@ static struct bpf_insn *bpf_convert_tstamp_read(const struct bpf_prog *prog,
/* AX is needed because src_reg and dst_reg could be the same */
__u8 tmp_reg = BPF_REG_AX;
- *insn++ = BPF_LDX_MEM(BPF_B, tmp_reg, skb_reg, PKT_VLAN_PRESENT_OFFSET);
+ *insn++ = BPF_LDX_MEM(BPF_B, tmp_reg, skb_reg, SKB_BF_MONO_TC_OFFSET);
*insn++ = BPF_ALU32_IMM(BPF_AND, tmp_reg,
TC_AT_INGRESS_MASK | SKB_MONO_DELIVERY_TIME_MASK);
*insn++ = BPF_JMP32_IMM(BPF_JNE, tmp_reg,
@@ -9267,14 +9267,14 @@ static struct bpf_insn *bpf_convert_tstamp_write(const struct bpf_prog *prog,
if (!prog->tstamp_type_access) {
__u8 tmp_reg = BPF_REG_AX;
- *insn++ = BPF_LDX_MEM(BPF_B, tmp_reg, skb_reg, PKT_VLAN_PRESENT_OFFSET);
+ *insn++ = BPF_LDX_MEM(BPF_B, tmp_reg, skb_reg, SKB_BF_MONO_TC_OFFSET);
/* Writing __sk_buff->tstamp as ingress, goto <clear> */
*insn++ = BPF_JMP32_IMM(BPF_JSET, tmp_reg, TC_AT_INGRESS_MASK, 1);
/* goto <store> */
*insn++ = BPF_JMP_A(2);
/* <clear>: mono_delivery_time */
*insn++ = BPF_ALU32_IMM(BPF_AND, tmp_reg, ~SKB_MONO_DELIVERY_TIME_MASK);
- *insn++ = BPF_STX_MEM(BPF_B, skb_reg, tmp_reg, PKT_VLAN_PRESENT_OFFSET);
+ *insn++ = BPF_STX_MEM(BPF_B, skb_reg, tmp_reg, SKB_BF_MONO_TC_OFFSET);
}
#endif
diff --git a/tools/testing/selftests/bpf/prog_tests/ctx_rewrite.c b/tools/testing/selftests/bpf/prog_tests/ctx_rewrite.c
index d5fe3d4b936c..ae7b6e50e405 100644
--- a/tools/testing/selftests/bpf/prog_tests/ctx_rewrite.c
+++ b/tools/testing/selftests/bpf/prog_tests/ctx_rewrite.c
@@ -68,17 +68,17 @@ static struct test_case test_cases[] = {
#if defined(__x86_64__) || defined(__aarch64__)
{
N(SCHED_CLS, struct __sk_buff, tstamp),
- .read = "r11 = *(u8 *)($ctx + sk_buff::__pkt_vlan_present_offset);"
+ .read = "r11 = *(u8 *)($ctx + sk_buff::__mono_tc_offset);"
"w11 &= 160;"
"if w11 != 0xa0 goto pc+2;"
"$dst = 0;"
"goto pc+1;"
"$dst = *(u64 *)($ctx + sk_buff::tstamp);",
- .write = "r11 = *(u8 *)($ctx + sk_buff::__pkt_vlan_present_offset);"
+ .write = "r11 = *(u8 *)($ctx + sk_buff::__mono_tc_offset);"
"if w11 & 0x80 goto pc+1;"
"goto pc+2;"
"w11 &= -33;"
- "*(u8 *)($ctx + sk_buff::__pkt_vlan_present_offset) = r11;"
+ "*(u8 *)($ctx + sk_buff::__mono_tc_offset) = r11;"
"*(u64 *)($ctx + sk_buff::tstamp) = $src;",
},
#endif
--
2.39.2
^ permalink raw reply related [flat|nested] 5+ messages in thread
* [PATCH bpf-next v2 2/3] net: skbuff: reorder bytes 2 and 3 of the bitfield
2023-03-21 1:41 [PATCH bpf-next v2 0/3] net: skbuff: skb bitfield compaction - bpf Jakub Kicinski
2023-03-21 1:41 ` [PATCH bpf-next v2 1/3] net: skbuff: rename __pkt_vlan_present_offset to __mono_tc_offset Jakub Kicinski
@ 2023-03-21 1:41 ` Jakub Kicinski
2023-03-21 1:41 ` [PATCH bpf-next v2 3/3] net: skbuff: move the fields BPF cares about directly next to the offset marker Jakub Kicinski
2023-03-21 5:20 ` [PATCH bpf-next v2 0/3] net: skbuff: skb bitfield compaction - bpf patchwork-bot+netdevbpf
3 siblings, 0 replies; 5+ messages in thread
From: Jakub Kicinski @ 2023-03-21 1:41 UTC (permalink / raw)
To: martin.lau
Cc: ast, daniel, bpf, netdev, davem, edumazet, pabeni, Jakub Kicinski
BPF needs to know the offsets of fields it tries to access.
Zero-length fields are added to make offsetof() work.
This unfortunately partitions the bitfield (fields across
the zero-length members can't be coalesced).
Reorder bytes 2 and 3, BPF needs to know the offset of fields
previously in byte 3 and some fields in byte 2 should really
be optional.
The two bytes are always in the same cacheline so it should
not matter.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
include/linux/skbuff.h | 20 ++++++++++----------
1 file changed, 10 insertions(+), 10 deletions(-)
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index 5a63878a4550..36d31e74db37 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -944,16 +944,6 @@ struct sk_buff {
__u8 ip_summed:2;
__u8 ooo_okay:1;
- __u8 l4_hash:1;
- __u8 sw_hash:1;
- __u8 wifi_acked_valid:1;
- __u8 wifi_acked:1;
- __u8 no_fcs:1;
- /* Indicates the inner headers are valid in the skbuff. */
- __u8 encapsulation:1;
- __u8 encap_hdr_csum:1;
- __u8 csum_valid:1;
-
/* private: */
__u8 __mono_tc_offset[0];
/* public: */
@@ -966,6 +956,16 @@ struct sk_buff {
__u8 tc_skip_classify:1;
__u8 tc_at_ingress:1; /* See TC_AT_INGRESS_MASK */
#endif
+
+ __u8 l4_hash:1;
+ __u8 sw_hash:1;
+ __u8 wifi_acked_valid:1;
+ __u8 wifi_acked:1;
+ __u8 no_fcs:1;
+ /* Indicates the inner headers are valid in the skbuff. */
+ __u8 encapsulation:1;
+ __u8 encap_hdr_csum:1;
+ __u8 csum_valid:1;
#ifdef CONFIG_IPV6_NDISC_NODETYPE
__u8 ndisc_nodetype:2;
#endif
--
2.39.2
^ permalink raw reply related [flat|nested] 5+ messages in thread
* [PATCH bpf-next v2 3/3] net: skbuff: move the fields BPF cares about directly next to the offset marker
2023-03-21 1:41 [PATCH bpf-next v2 0/3] net: skbuff: skb bitfield compaction - bpf Jakub Kicinski
2023-03-21 1:41 ` [PATCH bpf-next v2 1/3] net: skbuff: rename __pkt_vlan_present_offset to __mono_tc_offset Jakub Kicinski
2023-03-21 1:41 ` [PATCH bpf-next v2 2/3] net: skbuff: reorder bytes 2 and 3 of the bitfield Jakub Kicinski
@ 2023-03-21 1:41 ` Jakub Kicinski
2023-03-21 5:20 ` [PATCH bpf-next v2 0/3] net: skbuff: skb bitfield compaction - bpf patchwork-bot+netdevbpf
3 siblings, 0 replies; 5+ messages in thread
From: Jakub Kicinski @ 2023-03-21 1:41 UTC (permalink / raw)
To: martin.lau
Cc: ast, daniel, bpf, netdev, davem, edumazet, pabeni, Jakub Kicinski
To avoid more possible BPF dependencies with moving bitfields
around keep the fields BPF cares about right next to the offset
marker.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
--
v2: fix the assembly in tstamp test
---
include/linux/skbuff.h | 18 +++++++++---------
.../selftests/bpf/prog_tests/ctx_rewrite.c | 8 ++++----
2 files changed, 13 insertions(+), 13 deletions(-)
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index 36d31e74db37..6aeb0e7b9511 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -947,15 +947,15 @@ struct sk_buff {
/* private: */
__u8 __mono_tc_offset[0];
/* public: */
- __u8 remcsum_offload:1;
- __u8 csum_complete_sw:1;
- __u8 csum_level:2;
- __u8 dst_pending_confirm:1;
__u8 mono_delivery_time:1; /* See SKB_MONO_DELIVERY_TIME_MASK */
#ifdef CONFIG_NET_CLS_ACT
- __u8 tc_skip_classify:1;
__u8 tc_at_ingress:1; /* See TC_AT_INGRESS_MASK */
+ __u8 tc_skip_classify:1;
#endif
+ __u8 remcsum_offload:1;
+ __u8 csum_complete_sw:1;
+ __u8 csum_level:2;
+ __u8 dst_pending_confirm:1;
__u8 l4_hash:1;
__u8 sw_hash:1;
@@ -1072,11 +1072,11 @@ struct sk_buff {
* around, you also must adapt these constants.
*/
#ifdef __BIG_ENDIAN_BITFIELD
-#define TC_AT_INGRESS_MASK (1 << 0)
-#define SKB_MONO_DELIVERY_TIME_MASK (1 << 2)
+#define SKB_MONO_DELIVERY_TIME_MASK (1 << 7)
+#define TC_AT_INGRESS_MASK (1 << 6)
#else
-#define TC_AT_INGRESS_MASK (1 << 7)
-#define SKB_MONO_DELIVERY_TIME_MASK (1 << 5)
+#define SKB_MONO_DELIVERY_TIME_MASK (1 << 0)
+#define TC_AT_INGRESS_MASK (1 << 1)
#endif
#define SKB_BF_MONO_TC_OFFSET offsetof(struct sk_buff, __mono_tc_offset)
diff --git a/tools/testing/selftests/bpf/prog_tests/ctx_rewrite.c b/tools/testing/selftests/bpf/prog_tests/ctx_rewrite.c
index ae7b6e50e405..4951aa978f33 100644
--- a/tools/testing/selftests/bpf/prog_tests/ctx_rewrite.c
+++ b/tools/testing/selftests/bpf/prog_tests/ctx_rewrite.c
@@ -69,15 +69,15 @@ static struct test_case test_cases[] = {
{
N(SCHED_CLS, struct __sk_buff, tstamp),
.read = "r11 = *(u8 *)($ctx + sk_buff::__mono_tc_offset);"
- "w11 &= 160;"
- "if w11 != 0xa0 goto pc+2;"
+ "w11 &= 3;"
+ "if w11 != 0x3 goto pc+2;"
"$dst = 0;"
"goto pc+1;"
"$dst = *(u64 *)($ctx + sk_buff::tstamp);",
.write = "r11 = *(u8 *)($ctx + sk_buff::__mono_tc_offset);"
- "if w11 & 0x80 goto pc+1;"
+ "if w11 & 0x2 goto pc+1;"
"goto pc+2;"
- "w11 &= -33;"
+ "w11 &= -2;"
"*(u8 *)($ctx + sk_buff::__mono_tc_offset) = r11;"
"*(u64 *)($ctx + sk_buff::tstamp) = $src;",
},
--
2.39.2
^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: [PATCH bpf-next v2 0/3] net: skbuff: skb bitfield compaction - bpf
2023-03-21 1:41 [PATCH bpf-next v2 0/3] net: skbuff: skb bitfield compaction - bpf Jakub Kicinski
` (2 preceding siblings ...)
2023-03-21 1:41 ` [PATCH bpf-next v2 3/3] net: skbuff: move the fields BPF cares about directly next to the offset marker Jakub Kicinski
@ 2023-03-21 5:20 ` patchwork-bot+netdevbpf
3 siblings, 0 replies; 5+ messages in thread
From: patchwork-bot+netdevbpf @ 2023-03-21 5:20 UTC (permalink / raw)
To: Jakub Kicinski
Cc: martin.lau, ast, daniel, bpf, netdev, davem, edumazet, pabeni
Hello:
This series was applied to bpf/bpf-next.git (master)
by Martin KaFai Lau <martin.lau@kernel.org>:
On Mon, 20 Mar 2023 18:41:12 -0700 you wrote:
> I'm trying to make more of the sk_buff bits optional.
> Move the BPF-accessed bits a little - because they must
> be at coding-time-constant offsets they must precede any
> optional bit. While at it clean up the naming a bit.
>
> v1: https://lore.kernel.org/all/20230308003159.441580-1-kuba@kernel.org/
>
> [...]
Here is the summary with links:
- [bpf-next,v2,1/3] net: skbuff: rename __pkt_vlan_present_offset to __mono_tc_offset
https://git.kernel.org/bpf/bpf-next/c/04aae213e719
- [bpf-next,v2,2/3] net: skbuff: reorder bytes 2 and 3 of the bitfield
https://git.kernel.org/bpf/bpf-next/c/b94e032b7ad6
- [bpf-next,v2,3/3] net: skbuff: move the fields BPF cares about directly next to the offset marker
https://git.kernel.org/bpf/bpf-next/c/c0ba861117c3
You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2023-03-21 5:21 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2023-03-21 1:41 [PATCH bpf-next v2 0/3] net: skbuff: skb bitfield compaction - bpf Jakub Kicinski
2023-03-21 1:41 ` [PATCH bpf-next v2 1/3] net: skbuff: rename __pkt_vlan_present_offset to __mono_tc_offset Jakub Kicinski
2023-03-21 1:41 ` [PATCH bpf-next v2 2/3] net: skbuff: reorder bytes 2 and 3 of the bitfield Jakub Kicinski
2023-03-21 1:41 ` [PATCH bpf-next v2 3/3] net: skbuff: move the fields BPF cares about directly next to the offset marker Jakub Kicinski
2023-03-21 5:20 ` [PATCH bpf-next v2 0/3] net: skbuff: skb bitfield compaction - bpf patchwork-bot+netdevbpf
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).